HRMIM
Your Budget Is a Measurement Instrument: What Targets Do to Behaviour

Your Budget Is a Measurement Instrument: What Targets Do to Behaviour

Beykan Çizel · Academic· 23 September 2026· 14 min read

A revenue budget gets approved in October. The number goes on paper, the line locks, the file closes. From that moment it starts doing two separate jobs. The first is measurement: at year end you will compare the actual against this number. The second is production: the behaviour that generates that actual will now be shaped by an awareness of the number on paper.

The second job undermines the reliability of the first. Measurement methodology has a general name for this — the instrument affects the phenomenon it measures. When you are measuring temperature you can ignore the problem. When you are measuring a management team against a budget target you cannot, because the party that knows the target is the same party that produces the result.

This piece is not about building the budget correctly. It is about what the budget does to people once it exists. Goal setting, budgetary slack, budget gaming and styles of budget-based evaluation have a literature going back more than forty years. Part of that literature is consistent, part of it is not, and the inconsistent part is the most instructive. I mark both as I go.

The moment a target is set, the object of measurement changes

Before the budget is approved, occupancy is a fact about the market — demand, rate, channel mix and capacity in combination. After approval, the same number becomes a fact about a second thing as well: how the team responded to a figure.

This does not make the measurement worthless. It changes what is being measured, and the person reading the scorecard usually does not notice the change. If food and beverage cost ratio closes a point under budget, nothing in your data tells you which of two stories actually happened: did purchasing genuinely run better, or did a planned December stock replenishment slide into January?

The financial case for budgeting holds regardless of this behavioural layer — decision timing, variance visibility, cash discipline. The behavioural layer does not refute the case; it limits the measurement claim built on top of it. Running without a budget is worse. Treating the budget as a pure measurement instrument is a separate mistake.

The budget as measurement instrument: what it measures and what it produces

Looking at a budget as a psychometric instrument is not a strained analogy. Structurally every component of an instrument is present: it has items (account lines), a scale (currency and period), a scoring rule (variance), and an inference from score to judgement (good or bad performance).

Every question you would ask of an instrument applies here. Does it cover the whole construct? Does it return the same result on repeated application under the same conditions? Is the inference chain from score to decision defensible? This is exactly where the construct validity discussion transfers line-for-line onto budget rows.

But there is one difference, and it is the difference that matters. A personality inventory is not designed to steer the respondent's behaviour over the following twelve months. A budget is designed for precisely that. Which means a budget is a measurement instrument built to produce the behaviour it measures. That is not a defect — it is the definition — and it is why reading a budget variance as a clean performance indicator is methodologically indefensible.

The practical consequence: the same number is made to serve two purposes at once, planning and evaluation. One figure carrying both jobs is the main reason it degrades at both.

Goal difficulty and performance: the linear relationship and its limits

Goal-setting research sits among the most heavily replicated findings in the behavioural sciences. In their review of a thirty-five-year programme, Locke and Latham (2002) describe the relationship between goal difficulty and performance as a positive linear function: as the goal gets harder, performance rises.

The effect sizes need care, because there is a widespread citation error at this point. The 2002 paper is a review; the source carrying the meta-analyses is Locke and Latham's 1990 book. The values reported in the review for goal difficulty fall in the range d = .52–.82. So "Locke and Latham (2002) found d = .52–.82" is wrong; the accurate phrasing is "the 2002 review reports meta-analytic effect sizes of d = .52–.82, from Locke and Latham (1990)". The distinction looks like a footnote detail, and it changes what kind of evidence the citation actually carries.

The hospitality translation is direct. Telling a department head to hold last year's average and telling them to move meaningfully above it should produce a measurable performance difference. Setting budget targets equal to last year's actuals is a deliberate decision not to use the most robust finding this literature has.

The commitment condition: where the relationship reverses

Nobody claims the linear relationship runs without limit, and where the limit sits is the most operationally useful part of this literature. In Locke and Latham's review, performance levelling off or dropping is tied to two conditions: reaching the limits of ability, or a lapse in commitment to a highly difficult goal (2002, p. 706).

The second condition is far more common in hotels than the first. An ability limit is a real ceiling — capacity, headcount, season length. A lapse in commitment is a psychological event, and it happens at the budget table. Set a target twelve points above an occupancy level the property has never reached, and the team will not reject the number. They will accept it and then ignore it. From the outside, the target has been agreed. Inside, it is no longer a target — it is a formality.

The measurement consequence: once commitment lapses, the variance stops carrying information about performance. The minus twenty points you see at year end is not a measure of a bad operation or a bad market. It is a measure of distance from a number nobody took seriously.

As a working rule: a target being hard and a target being credible are two separate properties, and the first does not function without the second. The question to ask before a budget goes for sign-off is not "is this ambitious enough" but "who believes this number, and on what grounds".

The specificity effect: why "do your best" cannot be measured

The second robust finding in the same literature concerns specificity. Specific, difficult goals are associated with higher performance than "do your best" instructions; the range reported in the 2002 review is d = .42–.80, again attributed to Locke and Latham (1990).

From a measurement standpoint the reason is obvious. "Do your best" defines no scale. With no external reference point the individual supplies their own, and the reference they supply sits almost always below what was reachable. A specific number fixes the scale from outside.

In hotels this shows up most often on the cost side. "Keep costs under control" is not a budget target, it is a wish. "Hold food cost ratio at 30.5% this season" is a target. A measurable behavioural difference between the two is exactly what the literature predicts. (That ratio is illustrative only, not an industry benchmark — no industry average appears anywhere in this piece.)

Specificity also carries a cost, and honesty requires stating it: a specific target simultaneously defines what falls outside it. Put a number on food cost ratio and everything without a number — plate quality, portion consistency, waste — becomes relatively unconstrained. Specificity raises performance and narrows attention. Those are two faces of one mechanism, and you cannot take one and leave the other.

The propensity toward budgetary slack

Budgetary slack is the comfort margin deliberately buried in a budget: expenses estimated high, revenue estimated low. From a department head's position it is rational behaviour — the wider the margin, the lower the variance risk.

The founding text in this area is Merchant's (1985) study in Accounting, Organizations and Society, which sets out the hypotheses on the propensity to create budgetary slack. A caution is needed here. In the popular retelling of this literature that paper gets cited as the source that proved tight targets cause slack, and I could find nothing to support that framing: there is no accessible record of what the paper's empirical tests returned, and setting out a hypothesis is not the same as confirming it. So I cite Merchant here only as the origin of the hypotheses. If you see the sentence "Merchant proved budgetary slack", it is worth asking what it rests on.

For slack behaviour itself there is firmer ground. Jensen (2003) shows that lying in budget processes appears in two distinct places: in the formulation of budgets, where superiors and subordinates both distort the figures, and in the realisation of them, where the operation is gamed to make the outcome fit the number.

In hotels, slack hides most comfortably in maintenance and general administrative lines. Occupancy and ADR are externally visible, comparable and argued over. What a maintenance line genuinely needs is known only to the department that owns it. The row with the highest information asymmetry is the row with the highest slack.

Tying the target to a bonus: measurement or negotiation

The centre of Jensen's argument is that the source of the behaviour is not the budget but the pay structure. Target-linked bonuses build a system that pays people for lying. His proposed remedy is not to abandon budgeting — this distinction is worth underlining, because Jensen is regularly presented as an anti-budget source and he is not. What he proposes is removing the kinks from the pay-performance profile: making the bonus curve a straight line rather than a step that jumps at exactly the target.

What the step does is easy to see in a hotel. If the bonus switches on at 100%, two behaviours surface in the final quarter:

SituationBehaviour it producesHow it looks in the budget
Target looks unreachablePush revenue into the next period, bury this year on purposeWorse this year, a "recovery" next year
Target sits just out of reachBuy occupancy with discounting, sacrifice ADRTarget met, profitability down
Target comfortably clearedCarry remaining revenue and spend into the next periodThis year plateaus, next year's base rises

In all three cases the budget target can look met. In all three, what you measured is not performance but positioning relative to a bonus threshold.

The design lesson is clear enough. When the measurement function and the payment function are loaded onto the same number, the number becomes negotiable. Separating measurement from pay entirely is not realistic in most groups; removing the step from the bonus curve is realistic, and it is the direction the literature points.

Styles of budget-based evaluation and their behavioural consequences

How a manager uses the budget may matter more than the budget itself. The framing here comes from the three-way classification set out in Hopwood's 1973 book An Accounting System and Managerial Behaviour: budget-constrained, profit-conscious and non-accounting styles of evaluation. Under a budget-constrained style the manager evaluates a subordinate essentially on short-term budget variance; under a profit-conscious style budget data is one input into a judgement about long-run profitability; under a non-accounting style budget data carries little weight in evaluation at all.

Hopwood's 1972 empirical study in the Journal of Accounting Research found the budget-constrained style associated with higher job-related tension and damaged social relations. It is worth keeping the two apart: the classification belongs to the 1973 book, the tension finding to the 1972 paper. Most citation lists blur them together.

The hospitality version will be familiar. A general manager who evaluates a department head monthly by looking at one variance line is applying a budget-constrained style. Under that arrangement the head's first priority becomes keeping the monthly line clean — decisions get sequenced accordingly, and that sequence does not always face the same direction as the property's annual interest.

The finding that would not replicate: two answers to one question

Here we reach the most honest part of this literature, and my aim in this section is not to announce a result but to report a disagreement as it stands.

Otley (1978) repeated Hopwood's study in a different organisational context and could not reproduce the finding. More than that: contrary to Hopwood's hypothesis, he found the budget-constrained style associated with higher managerial performance. Two studies ask the same question and return different answers. The conflict is documented in detail in Noeverman's doctoral dissertation, published through Erasmus University (2008, pp. 25-27).

The literature has not settled this, and that needs saying without rounding. "Tight budget control produces tension" is not an established finding; it is one study's finding and another study's failure to reproduce it. In the management literature this situation is not unusual.

What it means in practice is that you cannot learn from the literature which of the two holds at your property. What you can ask is more modest and more useful: in our arrangement, what does close variance tracking actually produce — correction, tension, or slack? That is the same question as bad performance or a badly set target, asked one level up at the evaluation-style layer.

Context conditionality: why the result changes by property

Otley's own conclusion was that the difference between the two studies came from organisational context: the findings are conditional on it. That is an unsatisfying answer and most likely the correct one.

The context variables you would expect to move the effect of budget-based evaluation are abundantly present in hotels:

  • Interdependence between departments. The front office sells occupancy and in doing so sets the food and beverage cost base. If one department's variance depends on a neighbouring department's decision, charging that variance to that head is a measurement error.
  • Controllability. Energy unit price, exchange rates, channel commission — all visible on a budget line, none of them a function of the manager's decisions.
  • Level of uncertainty. In a city hotel demand is relatively predictable; in a seasonal resort one weather spell erases and rewrites the monthly variance.
  • Degree of participation. Who set the target? Participative budgeting has a literature of its own — Shields and Shields (1998) survey its antecedents — and a manager who helped set a target should be expected to respond to it differently.

Those four variables are enough to explain why the same budget discipline produces correction at one property and defensive behaviour at another. The inference: evaluation style cannot be standardised as group policy. What should be standardised is not the style but the definitions — which line is charged to whom, and what counts as controllable.

Single, multiple, composite — all three are problematic

An intuitive fix is usually proposed for everything above: do not look at budget variance alone, put other indicators alongside it. That proposal rests on weaker ground than people assume.

Ridgway (1956) sets out the dysfunctional consequences of performance measurement for all three forms. Single measures narrow behaviour and drive waste. Multiple measures create contradictory trade-offs and say nothing about which takes precedence. Composite measures produce tension along with role and value conflict.

Reading Ridgway as an advocate of multi-indicator systems is a common error, and the opposite is true: the argument of the paper is that all three forms are dysfunctional. Adding five more rows next to budget variance does not solve the problem, it changes its shape.

The conclusion is obviously not to stop measuring. The conclusion is not to expect the solution to come from the composition of the indicator set. It comes from how the variance is read: is the variance a verdict or a question? The same number, on the same row, can be used either way — and which one it is is a property of management practice, not of the instrument.

What this looks like in hospitality: seasonal targets and carried commitments

In a seasonal operation, the budget's claim to be a measurement instrument breaks in two places during the year.

The first is periodisation. An annual target cannot be split into equal monthly slices for a property that earns most of its revenue in five months. An April variance carries no performance information; what it carries is a reading of how crude your monthly allocation assumption was. An arrangement that reads off-season variance as a verdict puts the team on the defensive in the first quarter, and the defensive habit carries into the season.

The second is the carried commitment. A maintenance job deferred in October appears as a favourable variance on this year's expense line and an unfavourable one on next year's. The number moves twice, the work happens once. The same mechanism runs on the people side: training deferred at season end, a position left unfilled, leave not taken — each improves one year's scorecard and books a debt against the next.

A third break is the seasonal workforce itself. The target is set in October, the season starts in April, the evaluation happens in November. A significant share of the team the target was addressed to is not at the review table. Applying an instrument after the people who produced the behaviour it measures have left makes its output hard to use as a performance judgement.

None of these three invalidates the budget. Together they invalidate reading the variance on its own.

Tying target setting back to measurement discipline

The practical discipline that follows is not a new indicator set. It is a record kept alongside the budget line.

Write the assumption next to the target. A target is not a number; it is a number plus the assumptions that generate it — occupancy, ADR, unit price, headcount. If the assumption is not recorded, at year end you cannot tell which assumption the variance came from, and the variance starts talking about performance instead.

Mark controllability. One field beside every row: how much of the variance on this line is a function of this manager's decisions? If the answer is "partly", the row belongs in monitoring, not evaluation.

Ask about commitment separately. When the target is approved, one question to whoever signs: do you believe this number, and if not, which assumption is the problem? The written answer to that question determines how the variance gets read at year end.

Keep carried items visible. A deferred commitment is not a favourable variance; it is a relocated cost. If items moved between periods do not appear on a separate line, the budget comparison reads in the wrong direction for two years running.

One methodological note, offered as an analogy and no more. A budget variance is a difference score — actual minus target. Cronbach and Furby (1970) argue that raw change and gain scores "lead to fallacious conclusions, primarily because such scores are systematically related to any random error of measurement", and that such scores are rarely useful however they are adjusted. A budget variance and a psychometric gain score are not the same thing — a target is a decision, not a measurement, and that difference is not trivial. But the two share something: in both cases a difference derived from two components gets interpreted without knowing the uncertainty in the components. Accounting for how the target was built when you read the variance is therefore not fastidiousness. It is a requirement.

That discipline has one structural precondition: holding the target and the actual in one structure. If the target lives in one file, the assumption in another, and the actual in the accounting system, none of the four records above can be maintained — there is no common reference to compare against.

Conclusion

A budget is a measurement instrument built to produce the behaviour it measures. That sentence is not a criticism, it is the definition of the instrument — and once you accept the definition, how you read a variance changes.

The robust part of the literature says this: specific, difficult targets are positively associated with performance, and the relationship breaks when ability limits are reached or commitment lapses. When a target is fused to a bonus, the behaviour you get is positioning rather than measurement, and the fix is not removing the budget but removing the step from the bonus curve.

The part of the literature that is not consistent is at least as instructive. Whether a budget-constrained evaluation style produces tension or performance remains open between Hopwood and Otley to this day. That uncertainty is better read as an instruction than as a gap: you have to measure what happens at your own property, because the literature is not answering on your behalf.

What remains is small and belongs to record-keeping. Write the assumption next to the target. Mark how much of the variance is a function of the manager's decisions. Ask about commitment at the approval table and record the answer. Show items carried between periods separately. In an arrangement that does those four things the variance still will not be a performance measure — but it will be a number whose content is known, and for most of what reaches a scorecard that alone counts as progress.


If an outside read on how your budget targets get set, which assumptions they rest on and how they are read through the year would help, the HRMIM team works with hotel groups on measurement and target structures regularly. To talk through a single target line in your own budget, get in touch.

Get New Posts by Email

We will email you when a new blog post is published.

Loading the form…

Related Posts

Budget vs Actual: Does the Variance Measure Performance or the Budget?

Budget vs Actual: Does the Variance Measure Performance or the Budget?

When a line closes twenty percent under budget, the first instinct is to question the operation. But a variance is the difference between two things, and one of them is the budget itself. This piece treats the variance as a measurement problem: is it measuring performance, or the accuracy of the budget? The literature on budget-based evaluation carries a clear warning here — and the contradiction inside that literature shows how much weight one variance figure can carry.

4 November 2026 · 13 min read

Far Below Target: Bad Performance or a Badly Set Target?

Far Below Target: Bad Performance or a Badly Set Target?

When an indicator lands far below target, the first reflex is to question operations. Usually the wrong reflex. A deviation can point to three different things: a genuine operational problem, a target that was set wrong to begin with, or a definition that does not hold across properties. This piece builds a decision matrix that separates the three, then works through the conditions that make period-over-period comparison meaningful — stable definitions, stable weights, comparable periods.

15 July 2026 · 11 min read

Revenue Up, Turnover Down, Stock On Target: Scoring KPI Direction

Revenue Up, Turnover Down, Stock On Target: Scoring KPI Direction

Revenue above target is good. Turnover above target is bad. Stock days should sit close to target, in either direction. Three indicators, three scoring directions — and no single formula scores all three correctly. This piece works through target direction and the scoring methods that follow from it, with hospitality examples: actual/target, target/actual, higher-is-better, lower-is-better and closeness-to-target. Including why an indicator with no target should not be scored at all.

1 July 2026 · 12 min read

Let's work with HRMIM

Explore our platform free of charge.